• Journal of KIISE:Information Networking
• /
• v.31 no.6
• /
• pp.633-642
• /
• 2004

A self-adaptation with context-awareness is needed within network to meet costumed services according a user's changing constraints. But the existing network has many difficulty in adding new functions because of slow standardization of network and slow deployment of new services. To solve this problem, an active network can support the suitable environment to add new function such as self- adaptation. Therefore, this Paper suggests Self Adaptation Service(SAS) using agent-based active network and the constraint-based Service Creation Environment(SCE) to support self-adaptation with context-awareness. SAS provides benefits to support the context-aware service and the fast deployment of new services.

• Proceedings of the KSPS conference
• /
• 1996.10a
• /
• pp.464-467
• /
• 1996

This paper aims to exploit inter/intra-speaker phoneme sub-class variations as criteria for adaptation in a phoneme recognition system based on a novel neural network architecture. Using a subcluster neural network design based on the One-Class-in-One-Network (OCON) feed forward subnets, similar to those proposed by Kung (2) and Jou (1), joined by a common front-end layer. the idea is to adapt only the neurons within the common front-end layer of the network. Consequently resulting in an adaptation which can be concentrated primarily on the speakers vocal characteristics. Since the adaptation occurs in an area common to all classes, convergence on a single class will improve the recognition of the remaining classes in the network. Results show that adaptation towards a phoneme, in the vowel sub-class, for speakers MDABO and MWBTO Improve the recognition of remaining vowel sub-class phonemes from the same speaker

• Journal of the Korea Society of Computer and Information
• /
• v.9 no.3
• /
• pp.149-155
• /
• 2004

This paper describes the self-adaptive Content Service Network (CSN) on Application Level Active Network (ALAN). Web caching technology comprises Content Delivery Network (CDN) for content distribution as well as Content Service Network (CSN) for service distribution. The IETF working group on Open Pluggalble Edge Service (OPES) is the works closely related to CSN. But it can be expected that the self-adaptation in ubiquitous computing environment will be deployed. The existing content service on CSN lacks in considering self-adaptation. This results in inability of existing network to support the additional services. Therefore, in order to address the limitations of the existing networks, this paper suggests Self-adaptive Content Service Network (CSN) using the GME and the extended ALAN to insert intelligence into the existing network.

• Journal of Sensor Science and Technology
• /
• v.30 no.5
• /
• pp.290-294
• /
• 2021

Experts have designed popular and successful model architectures, which, however, were not the optimal option for different scenarios. Despite the remarkable performances achieved by deep neural networks, manually designed networks for classification tasks are the backbone of object detection. One major challenge is the ImageNet pre-training of the search space representation; moreover, the searched network incurs huge computational cost. Therefore, to overcome the obstacle of the pre-training process, we introduce a network adaptation technique using a pre-trained backbone model tested on ImageNet. The adaptation method can efficiently adapt the manually designed network on ImageNet to the new object-detection task. Neural architecture search (NAS) is adopted to adapt the architecture of the network. The adaptation is conducted on the MobileNetV2 network. The proposed NAS is tested using SSDLite detector. The results demonstrate increased performance compared to existing network architecture in terms of search cost, total number of adder arithmetics (Madds), and mean Average Precision(mAP). The total computational cost of the proposed NAS is much less than that of the State Of The Art (SOTA) NAS method.

• Journal of Institute of Control, Robotics and Systems
• /
• v.5 no.8
• /
• pp.898-906
• /
• 1999

A control structure is introduced for the purpose of rejecting periodic (or repetitive) disturbances on a tracking system. The objective of the proposed structure is to drive the output of the system to the reference input that will result in perfect following without any changing the inner configuration of the system. The structure includes an adaptation block which learns the dynamics of the periodic disturbance and forces the interferences, caused by disturbances, on the output of the system to be reduced. Since the control structure acquires the dynamics of the disturbance by on-line adaptation, it is possible to generate control signals that reject any slowly varying time-periodic disturbance provided that its amplitude is bounded. The artificial neural network is adopted as the adaptation block. The adaptation is done at an on-line process. For this , the real-time recurrent learning (RTRL) algoritnm is applied to the training of the artificial neural network.

• ETRI Journal
• /
• v.39 no.5
• /
• pp.695-706
• /
• 2017

This paper presents a caching algorithm that offers better reconstructed data quality to the requesters than a probabilistic caching scheme while maintaining comparable network performance. It decides whether an incoming data packet must be cached based on the dynamic caching probability, which is adjusted according to the priorities of content carried by the data packet, the uncertainty of content popularities, and the records of cache events in the router. The adaptation of caching probability depends on the priorities of content, the multiplication factor adaptation, and the addition factor adaptation. The multiplication factor adaptation is computed from an instantaneous cache-hit ratio, whereas the addition factor adaptation relies on a multiplication factor, popularities of requested contents, a cache-hit ratio, and a cache-miss ratio. We evaluate the performance of the caching algorithm by comparing it with previous caching schemes in network simulation. The simulation results indicate that our proposed caching algorithm surpasses previous schemes in terms of data quality and is comparable in terms of network performance.

• The Journal of Information Systems
• /
• v.27 no.2
• /
• pp.135-155
• /
• 2018

Purpose Recently, with the enhancement of mobile technologies, people have formed various relationships and spreaded networks on social network service(SNS). In addition, although people make a decision based on the thoughts and emotions about self, there is little empirical research on social relations and self-construal of users in social network game (SNG). Design/methodology/approach This study was designed to examine the structural relationships among SNG users' social capital, self-construal, behavioral adaptation, SNG reuse intention and charged item purchasing intention. Findings The results from this study are as follow. First of all, the bonding social capital did not have a significant impact on behavioral adaptation to SNG, but bridging social capital had a positive impact on behavioral adaptation. Second, independent self-construal did not have a significant impact on behavioral adaptation to SNG, but interdependent self-construal had a positive impact on behavioral adaptation. Lastly, the behavioral adaptation to SNG had a positive impact reuse intention and charged item purchasing intention. Also, SNG reuse intention had a positive impact on charged item purchasing intention.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.5
• /
• pp.765-776
• /
• 1995

This paper describes a speaker adaptation method to improve the recognition performance of MLP(multiLayer Perceptron) based HMM(Hidden Markov Model) speech recognizer. In this method, we use lst-order linear transformation network to fit data of a new speaker to the MLP. Transformation parameters are adjusted by back-propagating classification error to the transformation network while leaving the MLP classifier fixed. The recognition system is based on semicontinuous HMM's which use the MLP as a fuzzy vector quantizer. The experimental results show that rapid speaker adaptation resulting in high recognition performance can be accomplished by this method. Namely, for supervised adaptation, the error rate is signifecantly reduced from 9.2% for the baseline system to 5.6% after speaker adaptation. And for unsupervised adaptation, the error rate is reduced to 5.1%, without any information from new speakers.

• Journal of Korean Institute of Industrial Engineers
• /
• v.26 no.4
• /
• pp.306-314
• /
• 2000

In this paper we present a new approach to estimate link travel speed based on the hybrid neuro-fuzzy network. It combines the fuzzy ART algorithm for structure learning and the backpropagation algorithm for parameter adaptation. At first, the fuzzy ART algorithm partitions the input/output space using the training data set in order to construct initial neuro-fuzzy inference network. After the initial network topology is completed, a backpropagation learning scheme is applied to optimize parameters of fuzzy membership functions. An initial neuro-fuzzy network can be applicable to any other link where the probe car data are available. This can be realized by the network adaptation and add/modify module. In the network adaptation module, a CBR(Case-Based Reasoning) approach is used. Various experiments show that proposed methodology has better performance for estimating link travel speed comparing to the existing method.

• Korean Journal of Remote Sensing
• /
• v.38 no.2
• /
• pp.199-213
• /
• 2022

The unsupervised domain adaptation can solve the impractical issue of repeatedly collecting high-quality training data every year for annual crop classification. This study evaluates the applicability of deep learning-based unsupervised domain adaptation models for crop classification. Three unsupervised domain adaptation models including a deep adaptation network (DAN), a deep reconstruction-classification network, and a domain adversarial neural network (DANN) are quantitatively compared via a crop classification experiment using unmanned aerial vehicle images in Hapcheon-gun and Changnyeong-gun, the major garlic and onion cultivation areas in Korea. As source baseline and target baseline models, convolutional neural networks (CNNs) are additionally applied to evaluate the classification performance of the unsupervised domain adaptation models. The three unsupervised domain adaptation models outperformed the source baseline CNN, but the different classification performances were observed depending on the degree of inconsistency between data distributions in source and target images. The classification accuracy of DAN was higher than that of the other two models when the inconsistency between source and target images was low, whereas DANN has the best classification performance when the inconsistency between source and target images was high. Therefore, the extent to which data distributions of the source and target images match should be considered to select the best unsupervised domain adaptation model to generate reliable classification results.